LSE Research Proposal Example: Hydrology student to flood insurance policy (Score 93)

Flood insurance in England operates under an unusual institutional arrangement. The Flood Re scheme, introduced in 2016, pools high-risk residential properties into a subsidised reinsurance facility, effectively decoupling premium pricing from the underlying physical hazard for a large segment of the market. This arrangement was designed as a transitional mechanism, with a statutory obligation to phase out by 2039 and return properties to risk-reflective pricing. The practical question of how insurers will price residual flood risk once that subsidy is removed has received limited systematic attention, particularly from the perspective of what hydrological evidence can and cannot support at the property scale. This proposal investigates the following research questions. First, to what extent do current fluvial flood hazard datasets used in the English insurance market accurately represent property-level annual exceedance probabilities, and where are the principal sources of epistemic uncertainty? Second, how do insurers currently translate hazard estimates into underwriting decisions, and what assumptions bridge the gap between modelled inundation extent and actuarial loss expectancy? Third, what adjustments to hazard communication or data standards would be necessary to support risk-reflective pricing at scale after Flood Re's planned withdrawal? The rationale for these questions is practical and tractable. The 2039 transition deadline creates a defined policy window. Existing flood mapping infrastructure — principally the Environment Agency's Risk of Flooding from Rivers and Sea (RoFRS) dataset and associated depth-damage functions — was designed primarily for land-use planning rather than insurance underwriting. The mismatch between planning-grade spatial resolution and the property-level precision that actuarial pricing requires is documented in practitioner literature but has not been examined systematically through a research design that integrates hydrological uncertainty analysis with insurance market evidence. Two bodies of scholarship are directly relevant, and they rarely speak to each other with sufficient precision. Within flood hydrology, a substantial literature addresses uncertainty in flood frequency estimation, including parameter uncertainty in statistical flood models, structural uncertainty in hydraulic routing, and the sensitivity of inundation extents to digital elevation model resolution. Work in this tradition — associated broadly with researchers at institutions including Exeter, Bristol, and Delft — has refined ensemble approaches to inundation modelling and demonstrated that return-period estimates at ungauged or poorly gauged catchments carry confidence intervals wide enough to shift a property between low and high-risk categories. What this literature does not typically do is translate uncertainty bounds into insurance-relevant metrics such as probable maximum loss or average annual loss, because those translations require assumptions about vulnerability and exposure that hydrologists treat as outside scope. Within insurance economics and climate finance, a separate literature examines adverse selection, moral hazard, and the political economy of flood insurance subsidies. Researchers working on the Flood Re transition have modelled premium trajectories and household affordability under various phase-out scenarios, and there is a growing strand of work on the role of flood risk disclosure in property markets. However, this literature tends to treat the hazard estimate as a given input — a probability number supplied by a model — rather than as an uncertain quantity with a distribution that itself has implications for pricing stability and reserve adequacy. The gap this proposal addresses sits precisely at the junction: no published study, to my knowledge, has taken a sample of English residential properties, propagated hydrological uncertainty through to actuarially expressed loss metrics, and then examined how that uncertainty range compares with the precision assumptions embedded in current underwriting practice. Closing this gap would not resolve the political economy of the Flood Re transition, but it would provide a more honest empirical foundation for the technical side of that debate. The study is structured in two phases. Phase one is a hydrological uncertainty analysis applied to a purposively selected set of catchments in England. I propose to work with publicly available gauged flow records from the National River Flow Archive and the Environment Agency's LiDAR-derived terrain data to construct ensemble inundation models for three to five catchments that span a gradient of gauging density and urban exposure. Using a Monte Carlo framework applied to flood frequency curves, I will generate distributions of inundation extent and depth at the one-in-thirty, one-in-one-hundred, and one-in-two-hundred return periods — the thresholds most commonly referenced in insurance underwriting guidelines. The output of this phase is not a new flood map; it is a characterisation of how wide the uncertainty envelope is around existing maps at property scale, expressed in units that an actuary can interpret. Phase two uses semi-structured interviews with practitioners in the flood insurance and reinsurance sector to examine how hazard data currently enters underwriting decisions. I anticipate recruiting approximately twelve to fifteen participants drawn from insurers, reinsurers, loss adjusters, and flood risk consultancies, using purposive sampling to cover both technical modellers and underwriting decision-makers. Interview data will be analysed thematically, with the specific aim of identifying where practitioners acknowledge uncertainty in hazard inputs, how they currently manage it, and what data characteristics they would require to support more granular risk-reflective pricing. The two phases are designed to be mutually informing rather than sequential in a strict sense: the quantitative uncertainty bounds from phase one will structure the interview questions in phase two, and interview findings may identify catchment types or return periods where the uncertainty problem is most practically consequential. The hydrological data required for phase one — National River Flow Archive records, Environment Agency LiDAR tiles, and RoFRS shapefiles — are publicly available under open government licence. No proprietary datasets are required for the core analysis, which is a deliberate scope constraint. If access to higher-resolution commercial terrain data or insurer loss databases were available, they would sharpen the analysis, but the research question is answerable without them. Phase two raises standard qualitative research ethics considerations. Interviews with industry practitioners will require LSE Research Ethics Committee approval before recruitment begins. Participants will be recruited through professional networks and industry bodies such as the Chartered Insurance Institute and the Flood and Coastal Risk Management network; I do not assume any pre-existing access. Informed consent, anonymisation of employer identity where requested, and secure data storage will follow LSE's standard research ethics protocols. There is a realistic risk that recruitment proves slower than anticipated in a commercially sensitive sector; the contingency is to supplement interviews with analysis of publicly available regulatory submissions, such as Flood Re's quinquennial reviews and Financial Conduct Authority consultation responses, which provide a documentary record of how the industry has articulated its data needs. A provisional timeline allocates the first term to literature consolidation and catchment selection, the second term to hydrological modelling and ethics approval, the third term to interviews and integrated analysis, and the final period to writing. This is consistent with a one-year MSc research project, though the modelling component carries execution risk if catchment data quality proves lower than anticipated on initial inspection. The Grantham Research Institute on Climate Change and the Environment at LSE hosts active research on climate risk, insurance markets, and the economics of adaptation, which is directly relevant to the second phase of this project. The Department of Geography and Environment provides methods training in spatial analysis and qualitative research design that maps onto both phases. LSE's library access to insurance industry databases and regulatory filings would support the documentary analysis contingency described above. The project sits at the boundary of physical geography and environmental policy, which is precisely the disciplinary position the MSc Environment and Sustainability programme is designed to occupy. The methods training in the programme — particularly in environmental data analysis and policy-relevant research design — would allow me to develop the actuarial translation component of phase one more rigorously than my undergraduate hydrology training alone supports. I am not claiming a pre-arranged supervisory relationship; I am identifying the Grantham Institute's climate risk and insurance strand as the intellectual home most appropriate for this question, and I would seek to discuss supervision arrangements through the department's standard process.